I do that with cygwin. The few times I've done it the data has looked the way I expect to look. I haven't looked too deeply into it though. I am aware that can't just take the data without looking at it carefully. Since one has to do this with optimisation enabled and this aggregates the inline functions and optimises out some code it has to be looked at. But still, I've found it to be very useful. Vladimir Prus wrote:
Robert Ramey wrote:
Here's how I do it in the serialization library.
I have a directory boost/libs/serialization/performance which is very similar to the boost/libs/serialization/test directory. In it I have a couple of performance tests - I need many more. I run the normal bjam system with the profile switch set
Are you doing this on Linux? profiling options to gcc are fairly useless there, since they produce bogus data.
- Volodya
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost